Stumpy
Persistent AI agents that run 24/7 in your Slack, Telegram, SMS, or email
← Back to Docs
The MCP plugin lets your agent connect to external MCP servers, giving it access to tools and capabilities beyond what Stumpy provides built-in. If a service offers an MCP server, your agent can use it.
What It Does
MCP (Model Context Protocol) is a standard for connecting AI agents to external tools. With MCP enabled, your agent can connect to any MCP-compatible server and use the tools it provides.
For example, you might connect your agent to:
- A company's internal API exposed as an MCP server
- A specialized data service
- Custom tools you've built for your own workflow
- Third-party services that offer MCP endpoints
Enabling
- Go to your agent's page and click "+ Add Plugin" in the sidebar
- Find MCP and click Enable
Adding an MCP Server
On the MCP plugin page:
- Enter a name for the server (for your reference)
- Enter the server URL (the MCP endpoint)
- Choose the authentication method: None, API Key (with optional custom header and prefix), or OAuth
- Click Add Your agent immediately gains access to all tools provided by that server. You can connect to multiple MCP servers.
How Your Agent Uses External Tools
Once connected, the tools from external servers appear alongside your agent's built-in tools. Your agent uses them automatically based on what you ask.
For example, if you connect an MCP server that provides a "query_database" tool, you can ask your agent "look up the latest sales figures" and it will use that tool to fetch the data.
You do not need to tell your agent which tools to use. It sees the available tools and their descriptions and chooses the right one for the task.
MCP Servers vs. MCP Setup
These are two different things:
MCP Servers (this page) lets your agent connect to external MCP servers to use their tools. Your agent is the client, calling out to other servers.
MCP Setup lets you manage your Stumpy agents from MCP clients like Claude Code, Claude Desktop, or ChatGPT. You are the client, managing your agents from external tools.
They can be used together. Your agent can consume tools from external servers, and you can manage your agents from MCP clients.
Limitations
- The external MCP server must be accessible via HTTP
- Server credentials are encrypted before storage
- Tool availability depends on the external server being online and responsive
- Your agent trusts the tools provided by connected servers, so only connect to servers you trust
Похожие серверы
N.I.N.A. Advanced API
Control the N.I.N.A. (Nighttime Imaging 'N' Astronomy) software through its Advanced API.
Bitcoin MCP Server
MCP server dedicated to the Bitcoin ecosystem for traders, analysts, developers, and more.
Aegntic MCP Servers
A collection of Model Context Protocol (MCP) servers for various tasks and integrations, supporting both Python and Node.js environments.
Uber
Book Uber rides directly through your AI assistant.
Airthings Consumer
Monitor air quality with Airthings devices.
Ultra MCP SS
An MCP server for programmatic control of smartscreen.tv displays via HTTP and MCP commands, with YouTube integration.
Ingero
eBPF-based GPU causal observability agent with MCP server. Traces CUDA Runtime/Driver APIs via uprobes and host kernel events via tracepoints to build causal chains explaining GPU latency. 7 MCP tools for AI-assisted GPU debugging and root cause analysis. <2% overhead, production-safe.
GW2 MCP Server
Connects Large Language Models (LLMs) with Guild Wars 2 data sources. Requires a Guild Wars 2 API key for wallet functionality.
Cinode
Give agents a view of people, projects, skills and assignments
ScanMalware.com URL Scanner
MCP server for ScanMalware.com URL scanning, malware detection, and analysis